Nowflation CPI Gauge Today: 3.45% — How the Daily Inflation Index Is Built (October 2026)
Looking for inflation today? Nowflation Realtime is the headline measurement of prices being set today. This page explains the Nowflation CPI Gauge, which estimates what official CPI would read with its inputs refreshed today.
The Nowflation CPI Gauge reads 3.45% year over year as of 2026-10-18, against official CPI of 3.7% for Sep 2026 (NSA, quoted as BLS quotes it) — a gap of -0.25pp. It is a daily number, but only 39.7% of its basket weight is market-priced today: 4 of 21 components are priced by a live source (Shelter: Owned, Shelter: Rent, Motor Fuel, Used Vehicles), Shelter: Owned through the all-tenant rent index (new-lease rents, tenure-weighted, used as measured) that carries the owners'-equivalent-rent weight. The remaining components hold the Sep 2026 CPI print, released 2026-10-14, until the next release on 2026-11-10. Engine method v1.5; weights from the 2026 BLS relative-importance table.
The variants today
| Variant | YoY | Level (Jan 2018 = 100) | Live coverage | Shelter treatment | Weights |
|---|
| Nowflation CPI Gauge (CPI-comparable) | 3.45% | 136.65 | 39.7% | all-tenant rent index (v1.3): new-lease rents (Zillow ZORI 0.8 / Apartment List 0.2, chain-linked) averaged over the tenure of every lease in force — what the renting population pays this month, measured, not forecast — carrying both the rent weight and the owners’-equivalent weight | BLS CPI relative-importance table |
| Gauge (SA fuel, pace basis) (seasonally adjusted fuel) | 3.34% | 136.75 | 39.7% | same all-tenant rent index as the Gauge, with motor fuel seasonally adjusted by the BLS gasoline factors — the basis for the annualized 1- and 3-month pace | BLS CPI relative-importance table |
| Supercore (services ex-shelter) | 2.40% | 126.79 | 0.0% | excluded, along with every goods, food-at-home and energy component | BLS weights renormalized over transportation services, water/sewer/trash, medical care, education & communication and recreation (supercore v2, 2026-09-23; v1 also carried food away, personal care and the core residual) |
| PCE-weighted (out-of-pocket scope) | 3.43% | 134.68 | 19.2% | same all-tenant rent model as the Gauge | BEA PCE relative-importance shares (hand-seeded from Table 2.3.5) |
The pace basis reads 3.34% year over year against the Nowflation CPI Gauge's 3.45%. The difference is the seasonal adjustment of motor fuel: the December-chain weighting applies fuel's weight to its price change since December, and seasonal adjustment changes that change, not only the path within the year.
Every component on 2026-10-18
| Component | Weight | Level | YoY | MoM | BLS subindex YoY | Diff | Feed | Source codes | Data through | Age (days) | Contribution (bps) |
|---|
| Shelter: Owned (the headline prices this weight on the all-tenant rent index, the level, YoY and MoM shown; its own row, the official owners' equivalent rent subindex, reads 3.28% YoY, +0.35% MoM) | 26.2% | 148.2 | 1.97% | +0.18% | 3.28% | -1.30pp | LIVE (all-tenant rent index) | CUSR0000SEHC | 2026-09-30 | 18 | 51.6 |
| Medical Care | 8.4% | 124.25 | 1.50% | +0.12% | 1.50% | 0 | OFFICIAL | CPIMEDSL | 2026-09-01 | 47 | 12.6 |
| Food at Home | 8.3% | 134.39 | 1.77% | -0.07% | 1.77% | 0 | OFFICIAL | CUSR0000SAF11 | 2026-09-01 | 47 | 14.5 |
| Shelter: Rent | 7.8% | 148.2 | 1.97% | +0.18% | 2.78% | -0.81pp | LIVE | zori_us + aptlist_rent_us | 2026-09-30 | 18 | 15.4 |
| Transportation Services | 6.3% | 147.68 | 2.77% | +0.58% | 2.77% | 0 | OFFICIAL | CUSR0000SAS4 | 2026-09-01 | 47 | 17.6 |
| Education & Comm | 5.9% | 110.16 | 2.32% | +0.16% | 2.32% | 0 | OFFICIAL | CPIEDUSL | 2026-09-01 | 47 | 13.7 |
| Food Away | 5.4% | 147.07 | 3.46% | +0.23% | 3.46% | 0 | OFFICIAL | CUSR0000SEFV | 2026-09-01 | 47 | 18.8 |
| Recreation | 5.1% | 123.03 | 2.82% | +0.51% | 2.82% | 0 | OFFICIAL | CPIRECSL | 2026-09-01 | 47 | 14.5 |
| Household Furnishings | 4.3% | 129.65 | 1.55% | -0.05% | 1.55% | 0 | OFFICIAL | CUSR0000SAH3 | 2026-09-01 | 47 | 6.8 |
| New Vehicles | 3.8% | 122.58 | 0.61% | +0.20% | 0.61% | 0 | OFFICIAL | CUSR0000SETA01 | 2026-09-01 | 47 | 2.3 |
| Motor Fuel | 2.9% | 173.55 | 40.16% | -3.81% | 34.17% | +5.99pp | LIVE | aaa_gas_national | 2026-10-18 | 0 | 129 |
| Used Vehicles | 2.8% | 140.35 | -0.53% | -1.10% | -1.46% | +0.93pp | LIVE | manheim_muvvi | 2026-09-01 | 47 | -1.2 |
| Electricity | 2.5% | 143.9 | 4.37% | +0.25% | 4.37% | 0 | OFFICIAL | CUSR0000SEHF01 | 2026-09-01 | 47 | 10.5 |
| Personal Care | 2.5% | 133.63 | 3.72% | +0.28% | 3.72% | 0 | OFFICIAL | CUSR0000SAG1 | 2026-09-01 | 47 | 9.1 |
| Apparel | 2.4% | 109.48 | 3.37% | +0.31% | 3.37% | 0 | OFFICIAL | CPIAPPSL | 2026-09-01 | 47 | 8 |
| Lodging Away | 1.3% | 120.85 | 4.38% | +2.04% | 4.38% | 0 | OFFICIAL | CUSR0000SEHB | 2026-09-01 | 47 | 5.5 |
| Water, Sewer & Trash | 1.1% | 144.8 | 4.84% | +0.23% | 4.84% | 0 | OFFICIAL | CUSR0000SEHG | 2026-09-01 | 47 | 5.5 |
| Residual (core proxy) | 0.9% | 133.52 | 2.57% | +0.34% | 2.57% | 0 | OFFICIAL | CPILFESL | 2026-09-01 | 47 | 2.5 |
| Alcohol | 0.8% | 122.17 | 1.42% | +0.01% | 1.42% | 0 | OFFICIAL | CUSR0000SAF116 | 2026-09-01 | 47 | 1.2 |
| Utility Gas | 0.8% | 154.79 | 4.98% | -0.32% | 4.98% | 0 | OFFICIAL | CUSR0000SEHF02 | 2026-09-01 | 47 | 3.8 |
| Tobacco | 0.4% | 169.61 | 6.43% | +0.73% | 6.43% | 0 | OFFICIAL | CUSR0000SEGA | 2026-09-01 | 47 | 2.9 |
Rows: component method v1.5. YoY is the change each component enters the headline with (the owners'-equivalent weight rides the all-tenant rent index); Diff is that YoY minus the official subindex's. Contribution is each component's exact share of the headline's year-over-year change under the December-chain weighting; the column sums to 3.45% against the Nowflation CPI Gauge headline's 3.45%.
What fed the gauge today
- Shelter: Owned: priced on the all-tenant rent index, the Shelter: Rent row
- Shelter: Rent: Zillow Observed Rent Index (US, SA) (zori_us) — fresh, observation 2026-09-30 public since 2026-10-16, 18 days old, 80% of the blend; Apartment List national rent estimate (aptlist_rent_us) — fresh, observation 2026-09-01 public since 2026-10-02, 18 days old, 20% of the blend
- Motor Fuel: AAA national average regular (aaa_gas_national) — fresh, observation 2026-10-18 public since 2026-10-18, 0 days old, 100% of the blend
- Used Vehicles: Manheim Used Vehicle Value Index (manheim_muvvi) — fresh, observation 2026-09-01 public since 2026-10-07, 18 days old, 100% of the blend
- Medical Care, Food at Home, Transportation Services, Education & Comm, Food Away, Recreation, Household Furnishings, New Vehicles, Electricity, Personal Care, Apparel, Lodging Away, Water, Sewer & Trash, Residual (core proxy), Alcohol, Utility Gas, Tobacco: the Sep 2026 BLS subindex, released 2026-10-14. Electricity and utility gas are deliberately not overlaid (EIA residential prices publish 12–13 weeks after the month, later than the CPI subindex); food at home is held on purpose: the USDA advertised-price composite is ingested and graded as a diagnostic but not spliced, because it reads below the shelf price on average and worsened the print-eve grade at both horizons; the BLS subindex carries the component.
How the number is constructed
- Collect — every input is stored with a vintage date the day it is pulled: 20 BLS subindexes (via FRED, with their NSA twins) as the carry-forward history, a ~1% residual on core CPI, plus the live inputs (Zillow ZORI 0.8 / Apartment List 0.2 new-lease rents for the shelter model; AAA daily pump price for fuel; Manheim wholesale index for used vehicles, entering about a week after month-end when Cox publishes it; Kelley Blue Book / Cox average transaction price for new vehicles, monthly, what buyers actually paid). Nothing is back-filled from memory.
- Date by publication — a monthly print enters the daily history on the day it became knowable, never on its reference month: CPI subindexes step on the BLS release calendar, other monthly sources on their observed publication lag measured from the month-END they describe (ZORI about 16 days, Apartment List about 2 days, Manheim about 7 days; EIA weekly gasoline 2 days). The July CPI exists in the history from its August release date, so a year-ago comparison only sees what was public a year ago.
- Model shelter — new-lease rents measure the marginal tenant, whoever signs this month; the CPI re-prices each unit in its rent panel every six months, so its index measures every tenant, most of whom are mid-lease. The rent sources are chain-linked into one new-lease index, then averaged over the trailing 48 months with weights that decay at the monthly lease turnover (4.5% a month, a mean tenure of about 22 months) — what the CPI’s slowly rotating rent panel pays this month, used as measured (no forecasting map; the Gauge is a measurement). v1.4 lengthened the memory (from 6% / 36 months) and selects the parameters on the contemporaneous fit to CPI rent + OER, not on a lead; the index has read above the BLS all-tenant research index in recent quarters (ours minus BLS +0.69pp over its last three published quarters). The affine “where official rent lands” map (lead ~5 months, r 0.987) is a Lead-Test diagnostic only. Applied raw, new-lease rents measure the marginal tenant and never converged to the print.
- Rebase and splice — every series is indexed to January 2018 = 100. A live overlay is scaled to equal the official subindex on the day it starts (or re-starts), so a splice can never jump the level. Blends are chain-linked: each day’s move is the weighted average of the day-over-day moves of the sources that are fresh on both days, so a source dropping out or returning changes the composition without moving the level.
- Stale fallback — a live source stops counting once its newest data is older than its staleness window (14 days for daily feeds, 60 for monthly). The component then holds its last live level and moves with its official subindex from there: a dead feed can neither freeze the component nor snap it back to the official level. Coverage falls, the number keeps publishing.
- Official between prints — every component without a better daily source rides its BLS subindex at its last measured level until the next print lands (no extrapolation: the Gauge measures, the print-day model forecasts). Electricity and utility gas are deliberately official (EIA’s residential price series publish roughly 12–13 weeks after the month, later than the CPI subindex they would stand in for, and spot Henry Hub is not a regulated retail tariff); food at home stays official because the USDA advertised-price composite, though it correlates 0.71 with the CPI subindex, reads about 1.7pp below the shelf and worsened the print-eve grade at both horizons.
- Until September 27, 2026 (method v1.4): Aggregate (v1.4) — headline = Σ weight × component index (Laspeyres-style) with the published BLS relative-importance weights, the December table of the prior year governing each year, chain-linked at December 31 so the weight switch never moves the level (unchained, the true tables would have jumped the index +0.93% on 1 January 2022); the all-tenant rent index carries both the rent and the owners’-equivalent weight. Coverage = the share of basket weight fed by a live source that day; a day is only published when at least 95% of the weight has a value. Any live-source move over 5% in a day is held one day at the prior value.
- Until September 27, 2026 (method v1.4): Compare (v1.4) — YoY = today’s chained level ÷ the chained level a year ago composed from each component’s own base: the same calendar day 12 months earlier for a live overlay (clamped to the shorter month), and for a carried subindex the level 12 months before the print it holds — so a held print is always measured over exactly 12 months and matches the official subindex’s 12-month change to the basis point (a calendar-day lookup spanned 10, 11 or 13 months on 23% of print eves). MoM composes the same way. The official figure quoted next to the Gauge is the BLS YoY on the NSA index at one decimal, exactly as published; the Gauge itself is built on the seasonally adjusted subindexes.
- October 2025 — BLS collected no CPI for October 2025, so every subindex skips the month. For year-over-year bases only, each carried subindex’s October level is imputed as the geometric mean of its September and November prints, dated available on the November release (2025-12-18), so the vintage-true daily history is unchanged; the imputed base is in use for the October-2026 print (from its 10 November 2026 release until the November print on 10 December 2026) and is flagged on each component and in the QA self-test. April-2026 rent and OER carry a BLS imputation step (+0.55% in one month) for the panel that would have been collected in October 2025; it is treated as the measured print it is.
- Pace (v1.3) — the annualized 1- and 3-month pace is computed on the pace basis: the Gauge with motor fuel seasonally adjusted by the BLS gasoline factors, so a summer pump-price swing does not read as a change in trend. YoY is half last year’s prices; the pace is the change over the last one or three months at an annual rate. One month compounded twelve times is noisy: a pump-price jump or the day the held components step to a new CPI print moves it a lot, and the page names what moved it.
- From September 28, 2026 (method v1.5): BLS relative-importance aggregation (price relatives since December), monthly live sources compare the same month a year apart and step aside once the BLS has published a newer month; the index is linked to the v1.4 level on the day before the switch, and readings published before it are unchanged.
Engine v1.5 since 2026-09-28 (v1.4 before; readings published under it are unchanged): daily component indexes (base 2018-01 = 100) built from publication-dated BLS subindexes (held at their last measured print between releases; 21 components) with live overlays spliced on — an all-tenant rent index for shelter, daily pump prices, wholesale used-vehicle values, new-vehicle transaction prices — aggregated on the BLS December chain with the published relative-importance weights; YoY = the year-ago-to-December segment on the base year's weights times the December-to-today segment on this year's, each component on its own base (the held print's month−12 for carried subindexes, the source's own month a year earlier for a monthly live source, the calendar day a year earlier for a daily one); each component's contribution is its exact share of that change. Pace annualized on the seasonally adjusted variant, both ends on v1.5. Coverage counts the basket weight a live source prices that day. Every figure here is read from the engine's stored daily indexes and the records written at each run. Full methodology and changelog: methodology. Feed health: system status. Why the gauge and the print differ: why the gap.
Zillow data: Data Provided by Zillow Group.
Related: Why the gap · Methodology · Gauge vs BLS · The lead test